If a definitive genetic study of race and intelligence gets performed, what will the W/B genetic g difference be?

The g factor refers to general cognitive abilities; things that make you better or worse at a wide variety of cognitive tasks. Differences in the g factor are partly due to genetics, with estimates for its heritability ranging from 20% to 80%. Brain size has been implicated in g, IIRC accounting for like 10% of g.

There are various studious that investigate whether there are genetic differences in g between black people and white people. A lot of them are quite questionable because they are very indirect, looking at e.g. whether phenotypic differences after controlling for various correlated social factors.

In principle the most direct approach would be to just have a good genetic model of differences in g and applying that to genetic data that distinguishes the races. However, genetic models of psychological traits are still in their infancy, and we do not understand the mechanisms of genetic factors in much detail beyond their statistical average causal effects. This becomes a problem for various reasons that have been discussed in various places and which I won't go into in detail here.

Over time, presumably we will develop better models, to the point where eventually we will have a good mechanistic understanding of how g develops. At that point, presumably someone will do a definitive, solid study of the race differences in genetic g.

RESOLVES TO (white genetic g - black genetic g). To quantify genetic g, we will measure it with the within-race standard deviation of phenotypic g being 15. Mechanisms such as "black people have darker skin and society discriminates against darker-skinned people, giving them a lower phenotypic g" must be controlled for by the study. Mechanisms such as "black people know other black people and g is socially contagious, bringing the black g down due to historical disadvantage" or "black people have socioeconomic disadvantage in their families" may not confound the study in order for it to count for this market.

I will not be betting in this market.

Get Ṁ600 play money
Sort by:

Polygenic scores still explain only a timy fraction of the heritability we know about from twin studies etc. A lot of g variation is due to a long tail of rare mutations that aren’t amenable to inclusion in a PGS. How good do genetic models of g have to get to be used in the definitive study?

@JonathanRay It shouldn't just be a comparison of standard polygenic scores because polygenic scores are subject to various potential biases (e.g. the red-headed abuse model) and as you say, they don't capture all of the heritability.

It feels hard to say now ahead of time what exactly the rules should be because our understanding and methods will probably change over time to consider some paths more promising than others. One thing that would definitely be sufficient would be if we could come up with a good theoretically-justified mechanistic understanding for each racial genetic difference, and then counted up those differences. Theoretically-justified mechanistic understandings wouldn't have problems due to long tails of rare mutations as they aren't reliant on statistics over those mutations, but on the other hand it's not even clear that it will ever become possible to have this powerful of a theory. So I don't expect that kind of study to be the one that answers the question.

For quality's sake we should probably say that I will post the study for a month's worth of discussion on this market before resolving the market, so we have a chance of catching any serious problems.

Not sure if this sufficiently answers your question.

@tailcalled fair enough. Another issue is that a lot of environment is downstream of genetics. The environment of blacks living in the US is quite a lot better than that of blacks in any black countries. They reap large environmental benefits from living with a higher IQ population. Controlling for environment when environment is downstream of so much behavioral genetics is kind of an Everest regression. The US black-white gap is probably smaller than what you would get if you seeded 1000 identical exoplanets with 500 US white populations and 500 US black populations.

@JonathanRay Feel free to create your own market based on seeding 1000 exoplanets. Make sure to link it here.

@JonathanRay On the other hand, it may be well larger than what you would get from doing aggressive things to even out the environments. For instance if you did something crazy like CRISPRd everyone to have the same skin tone so no one could tell who is white and who is black and then had everyone randomly adopted, then I think the gap would decrease.

More related questions